Wolverine-Hawkeye Telecine

[David Brown]

Quick question for Stan. Can the Fan be run continuously, or does it need a switch for startup problems. That was true on version 7.
I could use another switch.

The AR0230 has some artifacts, I thought they came from capture or codec compression. It is a large and colorful image.

METS Telecine is impressive. He even made his own milling bit to cut the gate!!! Wow, I must take a minute to read thru his web site later.

I was thinking that projector claws are really thin and pointy. Look at the Wolverine claw. Sprocket hole tears I remember, are a thin cut in the bottom center. Easy enough to run the film upside down, but not if the tear cuts all the way thru to the next hole. This is how I imagine Kamel's film is torn.

This picture was cut with a knife for testing. Claw tears are not usually this long.



But what if the claw could be the same size and shape as the perforation? It could taper to a point from top to bottom and left to right. When it is fully inserted it could still hold the corners. Maybe just adding a "cap" to the Wolverine claw?

[Stan Jelavic]

Thank you Kamel. Actually the AR0230 does exhibit some pixel noise. Not sure what it is but maybe the sharpness setting. Sent a message to the arducam support. Hope they can clue me in to what the cause is.


The transport mechanism is complicated because it normally includes the gate. With this approach we do not need that gate. The Wolverine gate is used for image capture and this is just a "tractor as David put it" pulling the film. Here is a similar approach but we could have multiple claws in a row to reduce the chance of getting stuck due to bad perforations.
https://www.mets-telecinesystem.co.uk/film-transport/
This unit could be mounted where the pulleys are right now.

[Kamel Ikhlef]

Wow Stan !
This is the best rendering since the start of this adventure, Ithink.
I am very surprised given the price of this camera (better than my BUC02).
you can see it very clearly on the license plate. This will not increase the cost of the Hawkeye mods too much.
I am very far from getting your scan quality. (Just a little dark, but very good).
Well done Stan, I'm looking forward to the rest.

Thank you also for your idea of training damaged films, it's an idea to dig.

[Stan Jelavic]

Was just a thought. Multiple teeth are better than a single claw and the part with the teeth can be spring loaded laterally so that the precise adjustable stops can be added to match the frame width. The part does not include the gate making the design easier.
But anyways so much for that. I spent the day getting the AR0230 working with the Hawkeye and RPI. The issue with the max resolution git fixed by arducam which allowed me to run the capture at the max rez. Wrote a small python program that detects the trigger from the Hawkeye board and produces the sequence of images.
Turned out pretty good actually. Good dynamic range and pretty sharp. The colors are nice also.
https://drive.google.com/file/d/10EU...ew?usp=sharing

[David Brown]

I'm intrigued Stan! My first thought was it looks like a "tractor" pulling the film away. I saw a couple of youtube diy where just 3d parts and stepper motors advanced the film, no claw was used.

[Stan Jelavic]

Here is an idea...

[Kamel Ikhlef]

Thanks David for sharing your experience and testing, very helpful.
I haven't tested yet but can be printed soft filament TPU or filaflex for the smooth pulley would be a solution and we would keep the hollow shape to avoid scratching. To try as soon as I can...

For the moment I have not yet found an ideal solution for damaged films and I am thinking of an easier focusing solution with the small micrometer (still in reflection).

[Stan Jelavic]

Pretty thorough test David. Will check the high speed. Thought that that worked. Also if the sonalert is on continuously would mean that the motor is on continuously. Will test.

[David Brown]

I have been testing the capstan, here's what I found.

I thought my motor was flawed with some stiffness at 12-2pm. I ordered the new version. When I went to mount it, found I forgot to order screws!!!

My old motor is ok. I found the mount had put the motor shaft off center in the hole. With the added brass sleeve on the motor shaft there is still enough clearance, if it is centered. I drilled out the hole in the front panel to 3/16". FIXED.

If your not suffering from Capstan Fatigue, see my thoughts below on the toothed pulley.

I went thru the switch position chart to check operations. All is well.

EXCEPT, there is no alarm or capstan function when SPEED is switched high.

I put a switch on the motor to prevent the film from being moved when it stops. I tested the capstan with and without the motor active.

With the motor ON

ORIGINAL SONALERT: Chirps each time the capstan tries to pull the film, easily heard in the same room.

NEW SONALERT: Chirps each time the capstan tries to pull the film, easily heard in the next room.

With the motor OFF

ORIGINAL SONALERT: Sounds a high and low tone, pauses 1 second and repeats. The alarm is done. The stepper stops running and the take up motor continues to pulse every 20 seconds.

NEW SONALERT: Sounds a LOUD continuous tone forever. That means after 30 seconds you kill the power or your neighbors are knocking on your door asking, "YOU OK? The take up motor continues to pulse every 20 seconds.

I'm keeping both sonalerts. I'm adding a 3 position rocker for None / Low / High. Yes, I running out of space for switches!

Don't know why high/fast SPEED kills the capstan alarm? Still, it works with the others speeds.

The take up tension is working great. The capstan has tension on both sides and continues to rotate.

***************

Using the toothed pulley requires good precision that might be difficult to achieve. In this last round I discovered the motor mount was off center causing the shaft to rub on the face plate at 12-2pm and the back pulley had been drilled on tiny angle. Since then I found the motor mount or the motor's shaft is not completely perpendicular to the face plate. Then using a drill press to make a new back pulley, found it I had drilled off center. There is almost 1mm of eccentric orbit, caused by cumulative error from multiple parts.

Two solutions, 1- don't drill out the back pulley, fix the 3D model (that's a note to self). 2- don't use a toothed pulley!

To insure the capstan, using a smooth pulley, keeps turning, I made a change. I have eliminated the concave shape that is there to protect the film from scratching. I made it into a flat cylinder and added a rubber band. It is very soft and has great traction. Over time rubber bands go bad. I don't know where this one came from, it was in a box with all the M12 lenses I have. It might have been on a lens barrel of one?

This flat pulley can pull film with sprocket damage thru the gate from the second position. It's probably not reliable, I've been using the same test film each time. Sometimes it fails, but so does the toothed pulley.

I don't intend to force damaged film to continue, I'd rather have Hawkeye stop capturing and then deal with the problem.

I do like to have all options available. Most of all, I feel the capstan will not slip and cause a false alarm.

[Stan Jelavic]

It is AR0233 Bruce. Sorry.

[Bruce Davis]

Hi Stan, can not seem to be able to find a AR233 version wanted to look at specs.
thanks

[Stan Jelavic]

I am sidetracking the discussion a bit. Got the AR0230 camera working with the RPI. The image actually came out pretty good on my large monitor where my RPI is set up. Back on the laptop the shadows came out a bit dark.

The clouds are visible. The mid tones are good but on my laptop the shadows are a bit dark. Not the issue on the large 28" monitor.
There is an issue at full resolution the colors go crazy. This happens during image save. Sent a message to arducam support and hope they will point me in a right direction. Actually they were responsive and helped me with the driver problem that I had originally.
The lens shading with the 16mm lens seems to be acceptable.

Some intensity variation. The right hand side is dark because of the gate mask.


Still early to say but this sensor could be considered for HDR combined with a good quality monitor. The AR233 version might be OK also but there is some risk with shading around the edges since it is a larger sensor.

[Kamel Ikhlef]

Thanks David,
I knew the principle and had tried it a few years ago but it's so tedious.
This can be useful for sequences with damaged perforations.

Yes, I would like to find a powerful cropping software based on the perforation frame. To be continued...

[Stan Jelavic]

Good going David.

[David Brown]

Kamel, this might interest you.

It has been close to 10 years ago that I saw a DIY flatbed scanner project for digitizing 8mm film. I even bought an 8mm film holder for my scanner. This method is not for anyone with more than a couple of 3 inch reels! Even though multiple frames are digitized at one time, the film is advanced by hand.

It's still here, https://vimeo.com/22015773 and here https://wkurz.com/home/index.html


The author now has moved on to cameras. There are multiple executables for download. I think CineStripScanner is the one that creates single frames. I have not used the software and don't know if it could be used with single frame captures that need to be aligned. The author was available to user suggestions then, but it's been years!

This is the film holder I bought. I still have it and the scanner.

https://www.ebay.com/i/283618994018?chn=ps

A bit expensive, now I would design and print my own. I have since used it to hold the film on a light table, then shot single frames with my DSLR.

This process was so slow I abandoned it. The scanner is slow and the film must be advanced by hand. I never tried to process anything.

The point is 10-15 frames are scanned at once. Then they are automatically "cut out" and aligned in software using sprocket holes and other clues for alignment.

Just a thought, there are a few details to be considered.
- Modifying a Wolverine / Hawkeye or creating a simple light and camera with film supply and take-up.
-Triggering the camera without the claw. Each time the motor advances the film the camera could be triggered another way.
-The distance the film actually moves. Monitored by another sensor or a calculated. Small offsets would be corrected by software alignment and cumulative errors would stop the capture and the image corrected by hand, then restarted.

https://www.youtube.com/watch?v=AEPda-wZQTI

Maybe using a stepper motor (possibly geared) with a pinch roller, this could more accurately advance each frame for capture. The actual distances moved might not be as important if the image is over scanned. Eventually the process will drift too far and the capture stopped, the frame could be re-centered and restarted. A higher res stepper motor might last even longer, like a 0.9 degree.


If modifying a Wolverine or a projector, the claw should be disabled. The existing Wolverine stepper motor has gears that could be tapped into in two places to drive pulley.
The main gear is exposed in back. The small gear that drove the take up reel is still in place.


How this comes together is the hard part. Kamel, you have made great designs and have a printer!

For the rest of us, I don't know how much demand for this method there is.

Maybe there are other free frame alignment programs?



Stan my first thought was that the TURBO switch had failed. Turbo is now working all the time.

Testing with RUN=on SPEED=low TURBO=on I have 0 volts. With all switches off, I have 3.3 volts.

I have the new motor and once it's installed and I have the transport working, I'll move back to software.